[Mesa-dev] [PATCH 2/3] i965/fs: Make emit_if_gen6 never fall back to emit_bool_to_cond_code.

2014-09-04 Thread Kenneth Graunke
Matt and I believe that Sandybridge actually uses 0x for a true comparison result, similar to Ivybridge. This matches the internal documentation, and empirical results, but contradicts the PRM. So, the comment is inaccurate, and we can actually just handle these directly without ever

[Mesa-dev] [PATCH 2/3] i965/fs: Make emit_if_gen6 never fall back to emit_bool_to_cond_code.

2014-09-04 Thread Kenneth Graunke
Matt and I believe that Sandybridge actually uses 0x for a true comparison result, similar to Ivybridge. This matches the internal documentation, and empirical results, but contradicts the PRM. So, the comment is inaccurate, and we can actually just handle these directly without ever

Re: [Mesa-dev] [PATCH 2/3] i965/fs: Make emit_if_gen6 never fall back to emit_bool_to_cond_code.

2014-09-04 Thread Matt Turner
On Thu, Sep 4, 2014 at 12:18 AM, Kenneth Graunke kenn...@whitecape.org wrote: Matt and I believe that Sandybridge actually uses 0x for a true comparison result, similar to Ivybridge. This matches the internal documentation, and empirical results, but contradicts the PRM. So, the